18 ART. 2.-BASHF0RD DEAN. 



The wide range in the variational characters in species of 

 myxinoids has long been appreciated, in the matter, for example, of 

 the number of gills and ' teeth ' and in the proportions of body regions 

 and fins. My own observations lead me to the belief that in the 

 case of myxinoids it is peculiarly necessary to base specific deter- 

 minations upon the average characters of as great a number of 

 individuals as practicable. Unfortunately, but few specimens of the 

 southern species have as yet been examined critically, and until the 

 contrary is proven, we have, therefore, to accept their characters 

 already described as representing the average for these species. As 

 far as my own observations go, there is the greatest relative constancy 

 in the position of the outer branchial openings, i.e., in respect to 

 their location dorsal or ventral, anteriorly and posteriorly, and 

 as to the interspaces of the openings in each row. The relative 

 length of the external portion of the branchial duct in the 

 last as distinguished from the first gill pouches is also found 

 to be notably constant. Dental characters present somewhat 

 wide variations, although as far as numerical variations go, it 

 is not difficult to obtain the normal for each species. In such a 

 calculation the most troublesome question is to what degree immature 

 ' teeth ' should be counted : — or rather, the minimum size of those 

 which must be taken into account. For one often finds at the 

 posterior end of each row of ' teeth ' several more or less minute 

 cusps, which strictly speaking can be counted as ' teeth,' although 

 not actually functional. And by reckoning or not reckoning these 

 one would easily obtain the wide variations in a dental formula 

 which Ayers, for example, has given for H. stouti. The fusion of the 

 two or three cusps at the median end of each row of teeth appears to 

 be moderately constant for the species. It certainly has no value, 

 however, as a generic character, for we find that the same degree of 



